Академический Документы
Профессиональный Документы
Культура Документы
[CI] = 1.03–1.30) and borderline personality symptoms (OR = 1.07, 95% CI = 1.01–1.13)
were significantly associated with attempted suicide.
Conclusions
Our findings indicate that known sociodemographic risk factors for suicide may not apply
within psychiatric populations. Prevention strategies for suicidal behaviour in psychiatric
patients may be effective, including limited access to means for suicide attempts (i.e. excess
pharmaceutical drugs) and target screening for high-risk personality and impulsivity traits.
Introduction
Suicidal behaviours are complex and can be challenging to foresee even among patients
receiving medical and psychiatric care [1, 2]. Suicide is the second leading cause of death
among 15–29 year olds worldwide [3], with an even greater prevalence of non-fatal suicidal
behaviour [4]. Attempted suicide, defined as self-harm behaviour with intent to die [5], may
occur up to 20 times more frequently than completed suicide [4, 6]. Attempted suicide is
associated with adverse, long-term outcomes, including psychiatric and medical comorbid-
ity, hospitalization, repeated suicide attempts, poverty, chronic stress, and stigma [7, 8].
Considering the personal and public health burden of suicide on global and local scales, it is
necessary that preventative and rehabilitative strategies be developed to manage those pre-
senting with suicidal behaviour.
While individuals with psychiatric illnesses represent the vast majority of the individuals
who attempt suicide [4], only a small proportion of those with psychiatric illnesses attempt sui-
cide. Known risk factors for suicidal behaviours are largely based on studies of general com-
munity populations and these factors include prior suicide attempts, underlying psychiatric
and substance use disorders, single marital status, unemployment, and major life stressors [8–
12]. However, reliable predictors of suicidal behaviour among populations with serious psychi-
atric disorders remain elusive. Wide-scale screening of psychiatric patients has been suggested
as a method of early detection of suicidal behaviour [13], although feasibility greatly limits the
ability to comprehensively screen all patients. Some studies have examined suicidal risk factors
among patients with specific psychiatric disorders [14–16], yet suicidal risk factors among
broad psychiatric populations who typically present to clinical settings, including patients with
multiple psychiatric diagnoses, are not yet clearly established. Defining high-risk psychiatric
patients will allow clinicians to effectively screen patients for suicidal behaviour.
Moreover, given the difficulties of identifying those at risk of suicide from populations of
psychiatric patients, identification of warning signs and behaviours associated with suicide
among psychiatric patients can aid in suicide prevention. Patterns and behaviours associated
with suicide attempts are important to characterize within psychiatric populations, in order to
distinguish individuals in this group who attempt suicide from those with no history of suicide
attempts. Identifying trends in behaviours and methods of suicidal attempts within a sample of
patients with psychiatric disorders will aid in the movement towards developing large-scale
suicide prevention methods in clinical settings.
The present case-control study aimed to 1) describe the trends and circumstances associ-
ated with suicide attempts and 2) investigate risk factors of suicide attempts among adult inpa-
tients with psychiatric disorders.
Methods
The participants and data used in this investigation were collected for the Study of Determi-
nants of Suicide Conventional and Emergent Risk (DISCOVER) [17], a case-control study
designed to investigate risk factors of attempted suicide. Participants were recruited between
March 2011 and November 2014 in Hamilton, Ontario, Canada. Data were collected at
St. Joseph’s Healthcare and Hamilton Health Sciences hospitals. The study procedures were
approved by the Hamilton Integrated Research Ethics Boards (HIREB) (REB number 10–661
for St. Joseph’s Healthcare Hamilton and 11–3479 for Hamilton Health Sciences Hospitals).
We included adults ( 18 years of age) who were able to provide written informed consent,
communicate in English, and who were willing to follow study procedures. Fig 1 defines case
and control groups and outlines the recruitment process. Cases included psychiatric inpatients
who had made a suicide attempt, defined as self-directed injury with specific intent to die
and that necessitated admission to a medical or psychiatric hospital ward. We had initially
intended to recruit psychiatric inpatients who had made a suicide attempt within three months
of recruitment (case-recent group). However, given the challenges of recruiting this particular
patient population, we also included psychiatric inpatients who had a lifetime history of
attempted suicide (case-past group). The control group consisted of adult psychiatric inpatients
who had never attempted suicide and who were admitted to the same psychiatric hospital
within the same time frame as the cases. The predefined study design included matching of
case and control participants by sex and age (±5 years); however given the difficulties in recruit-
ing case participants, matching was not implemented in order to reach our target sample size.
Clinical staff identified eligible hospitalized patients who had the mental capacity to provide
written informed consent. These patients were consecutively approached by trained research
staff who inquired about their interest in participating in the study and provided study infor-
mation. Patients who agreed to participate were asked to read and sign consent forms describ-
ing study procedures. The local institutional ethics boards approved consent forms and
procedures. Research assistants conducted a structured face-to-face interview consisting of val-
idated questionnaires, described below. Participants were asked about sociodemographic char-
acteristics including age, sex, education and socioeconomic status. We assessed intention to
die as a result of the suicide attempt by asking the participants directly and measured level of
intent in relation to participants’ most recent suicide attempt using the Pierce Suicide Intent
Scale (P-SIS) [18]. The P-SIS consists of 12 questions assessing the circumstances during the
suicide attempt, self-reported risk, and medical risk. The domain scores are tabulated to pro-
vide an overall assessment of level of intent to die as a result of a suicide attempt. This scale dis-
tinguishes self-harm behaviour from suicide attempts. The overall score on the P-SIS ranges
from 0 to 25, with a maximum 3 or 4 points for each question, and higher scores correspond-
ing to higher intent to die. Total scores of 0–3 represent low intent, 4–10 represent moderate
intent, and 10 or more represent high intent to die. The scale has shown high test-retest reli-
ability (r = 0.97).
Participants also completed the 30-item Barratt Impulsiveness Scale (BIS) [19] to assess
trait impulsivity and the 23-item Borderline Symptom List (BSL) [20] to assess borderline per-
sonality symptoms. The BIS questionnaire was chosen to assess impulsivity as a risk factor
independent of psychiatric diagnoses. The BIS measures impulsive behaviours on attentional,
motor and nonplanning factors. Each question asks about an impulsive trait or behaviour on a
4-point Likert scale (Rarely/Never, Occasionally, Often, Almost Always/Always). The ques-
tionnaire has an overall maximum score of 120 and higher scores represent higher impulsivity,
such that an increase of 4 points is indicative of an additional impulsive trait or behaviour. The
questionnaire has shown good internal consistency (Cronbach’s α = 0.83) [19].
The BSL was used as a measure of borderline personality symptoms, with higher scores on
representing increased severity of symptoms. Each question asks about a borderline personal-
ity symptom on a 5-point Likert scale (Not at all = 0, A little = 1, Rather = 2, Much = 3, Very
strong = 4). The overall score is determined by dividing the sum of the individual item scores
with the total number of items (overall score = sum/23 for this study). Although there is no
clinical cut-off for diagnosis of borderline personality disorder on the BSL, researchers found a
mean score of 2.05 (SD = 0.90) in a sample of borderline personality disorder patients. For the
purposes of this study, we determined and reported the mean overall score for the case and
control groups as described above, however we used the total sum of the individual questions
as a continuous variable for the multivariable regression model (maximum score = 92). There-
fore, a 4-point increase represented an additional borderline personality symptom. The ques-
tionnaire has shown good internal consistency (Cronbach’s α = 0.97) and test-retest reliability
(r = 0.82, p<0.0001).
The Mini International Neuropsychiatric Interview (M.I.N.I.) [21] was administered to
determine whether participants met criteria for DSM-IV Axis I psychiatric disorders. We used
the M.I.N.I. to determine if participants had existing diagnoses of (1) mood disorders (major
depressive disorder, bipolar disorder), (2) anxiety disorders (generalized anxiety disorder,
panic disorder, social phobia, obsessive-compulsive disorder, post-traumatic stress disorder),
(3) substance use disorders (alcohol dependence/abuse and substance dependence/abuse), and
(4) psychotic disorders. The M.I.N.I. also determined if participants met the criteria for antiso-
cial personality disorder. For this study, an experienced psychiatrist determined the partici-
pants’ primary Axis I psychiatric diagnosis using the hierarchical rules in the DSM-III [22]
that were carried through to the DSM-IV classification system [23]. Therefore, the primary
diagnosis was assigned using the following hierarchy: (1) substance use disorder, (2) psychotic
disorder, (3) mood disorder, and (4) anxiety disorder.
STATA version 13 was used to perform all statistical analyses. For univariate analyses, we
used independent sample t-tests to compare means of continuous variables and chi-square
tests to compare proportions of categorical variables between cases and controls. Non-
parametric equivalents (i.e. Mann-Whitney-U tests) were used for continuous variables that
were not normally distributed. Fisher’s exact test was used to compare categorical variables
that had an expected frequency of less than 5 in a particular cell. Simple Pearson’s correlations
were used to assess the linear relationship between two normally distributed variables and
Spearman correlations were used for non-normally distributed variables. Multivariable logistic
regression analysis was utilized to assess clinical risk factors associated with suicidal attempts,
by comparing cases to controls. The Hosmer-Lemeshow test was used to assess the goodness-
of-fit of the regression model. Multi-collinearity between independent variables was assessed
using the variance inflation factor (VIF), and variables with VIF>10 were considered for
exclusion from the model. The level of significance was set at alpha = 0.05, and we included
clinically important variables based on the literature regarding psychiatric populations in the
logistic regression model [14–16, 24]. As a sensitivity analysis, we conducted multiple imputa-
tion using chained equations (MICE) to adjust for missing data in the multivariable regression
model [25]. Age and sex were used to aid in the prediction of missing values in the imputed
datasets. The reporting of this study is in accordance with the Strengthening of Reporting of
Observational Studies in Epidemiology (STROBE) guidelines [26].
Results
Study sample characteristics
The final sample comprised 250 individuals, including 146 psychiatric inpatients who had
attempted suicide (cases) and 104 psychiatric inpatients who had never attempted suicide
(controls). Fig 1 displays the number of individuals approached for recruitment and
included in the final sample as well as the reasons for exclusion. The sociodemographic
characteristics of the case and control groups are summarized in Table 1. The mean age of
https://doi.org/10.1371/journal.pone.0192998.t001
the case group was 45.18 years (standard deviation [SD] = 14.69 years, range 18–73 years)
and the control group was 45.01 years (SD = 14.23 years, range 18–82 years). The sample
consisted of an approximately equal proportion of males and females (55.48% females in
the case group, 50.00% females in the control group). In the univariate analyses, there
were no significant differences in sociodemographic factors between cases and controls
https://doi.org/10.1371/journal.pone.0192998.t002
(p-values >0.05). The case group scored significantly higher on both BSL (p = 0.0003) and
BIS (p = 0.0001) personality measures.
Psychiatric diagnoses were assessed using the M.I.N.I for 211 participants and we reviewed
medical records of 15 participants with missing M.I.N.I data to determine the psychiatric diag-
noses. Information on the primary psychiatric diagnosis was missing for 24 participants.
Table 2 presents the primary psychiatric diagnoses for the case and control groups. We utilized
the primary psychiatric diagnosis, however 136 participants (60.17%) had a psychiatric comor-
bidity. Additionally, there were 20 participants in the case group and no participants in the
psychiatric control group with a diagnosis of antisocial personality disorder although this was
none of the participants’ primary diagnosis. Univariate analyses showed no differences in the
prevalence of mood disorders, anxiety disorders, psychotic disorders or substance use disor-
ders between the case and control groups.
https://doi.org/10.1371/journal.pone.0192998.t003
Table 4. (Continued)
Notes: ‘N’ represents the total number of participants who responded to the question and ‘n’ represents the
proportion of participants with the specified response.
https://doi.org/10.1371/journal.pone.0192998.t004
or explosives (2/146). There were 14 participants who reported using multiple methods during
the suicide attempt.
The P-SIS assessed suicidal intent associated with the attempt based on circumstantial
factors (e.g. alcohol consumption, suicide note), self-reported factors (e.g. patient’s beliefs
regarding the lethality of the attempt), and medical risk (e.g. objective likelihood of death)
(Table 4). According to overall P-SIS scores, 80.18% of the individuals who attempted suicide
did so with high intent to die. Furthermore, 62.70% of the participants who attempted suicide
believed that they would have died from the attempt and more than half of the individuals
were found to have high medical risk for death based on objective questions. Of these indi-
viduals who attempted suicide, 47.69% reported feeling glad that they had recovered since
the attempt.
Table 5. Risk factors for attempted suicide: Multivariable logistic regression results (n = 211).
https://doi.org/10.1371/journal.pone.0192998.t005
Discussion
We sought to summarize characteristics and behaviours of adults associated with suicide
attempts, which may help classify factors influencing suicidal behaviour within a clinical set-
ting. We further aimed to make comparisons between individuals who had attempted suicide
and a control group of psychiatric inpatients with no history of suicide attempts to identify
risk factors among a vulnerable population. The findings indicated that adult psychiatric
patients who had attempted suicide did not significantly differ from the psychiatric control
group on sociodemographic characteristics or the prevalence of psychiatric diagnoses. The
case group showed higher prevalence of maladaptive personality measures such as impulsivity,
borderline symptoms and diagnosis of antisocial personality disorder. Factors such as single
marital status, unemployment and low education level did not significantly differ between the
case and control groups in this study, but have been reported as significant risk factors in stud-
ies comparing cases to community controls [4, 8, 9]. These findings suggest that known risk
factors of suicidal behaviour may not be applicable within psychiatric inpatient populations,
further confirming the necessity of identifying risk factors within psychiatric patients.
suicides in states with high gun ownership, while non-firearm related suicides were equal
across the states [30]. Similar results have been found in studies of adolescent suicides examin-
ing access to firearms in the household [32, 33]. An analogous trend may follow among psychi-
atric populations with access to high doses of pharmaceutical drugs. Furthermore, research in
the United Kingdom exploring limited access to over-the-counter pain relievers found that
changing legislation to reduce pack sizes and enforcing a limit in purchasable tablets led to sig-
nificant decreases in documented overdoses of these drugs over the following years [28]. These
strategies may be important to consider among Canadian psychiatric populations as a recent
study in Toronto found that among cases of completed suicides by overdose, prescription
medications (including opioid analgesics and psychotropic drugs) were involved in the major-
ity of suicides [34].
Evaluation of behaviours related to the suicide attempt revealed that almost half (45.03%)
of the individuals attempting suicide had somebody “present” and “nearby or in contact” at
the time of the suicide attempt. Furthermore, more than half (52.34%) the individuals who
attempted suicide took no precautions against discovery. This demonstrates the importance of
educating the families and friends of at-risk psychiatric patients regarding warning signs and
communication strategies. However, identification of warning signs remains challenging
given that a large proportion (80.00%) of individuals attempting suicide took no final actions
in anticipation of death (i.e. settling of affairs), which has been previously reported as a warn-
ing sign for suicidal behaviour [13]. Approximately one-quarter of the participants who had
attempted suicide left a note. Studies of individuals who completed suicide report a widely
varying proportion of suicide notes from 18% to 37% and show that youth are more likely to
leave notes [35].
The current study sample also showed that most cases of suicide attempts had serious
intent to die and the majority of patients believed that the attempt would lead to death. How-
ever, the proportion of individuals attempting suicide with high intent to die in this sample
appears greater than other samples, in which half of the participants report that they had low
intent to die or attempted as a “cry for help” [10]. This finding may be related to the higher
severity of psychiatric illness in this sample, since all participants were psychiatric inpatients.
While the majority of participants reported high intent to die at the time of attempt, 47.69%
of patients who had attempted suicide stated they were “glad that they recovered”. Given that
a history of attempted suicide is among the strongest predictors of completed suicide [3], it is
important to conduct future research among individuals reporting ambivalence related to
physical and psychological recovery, as this may play a role in the prevention of repeat sui-
cide attempts.
these studies were conducted among patients with specific psychiatric disorders, such as major
depression [39] or bipolar disorder [40].
Interestingly, trait impulsivity was associated with increased odds of attempted suicide yet
no significant correlation was found between trait impulsivity and the self-reported impulsive-
ness of the suicide attempt. These findings suggest that trait impulsivity may not necessarily
drive individuals to make impulsive suicide attempts. Baca-Garcia et al. found similar results
when exploring the relationship between trait impulsivity and suicide attempt impulsivity
[41]. This may have clinical implications in increasing the feasibility of suicide attempt predic-
tion among psychiatric patients, given the challenges associated with prevention of impulsive
attempts. Additionally, these results suggest that development of clinical screening tools for
trait impulsivity and borderline personality symptoms among psychiatric patients may be
important in suicide attempt prevention.
Limitations
This study was limited by potential biases due to self-report measures as well as the case-
control study design. Within the case group, the case-past group may have been differen-
tially affected by recall bias related to the details of the suicide attempt. Furthermore, the
cross-sectional design of the study did not allow us to determine the direction of the associa-
tion risk factors and suicide attempts. For example, suicide attempts, particularly in the
case-past group, may have occurred before the manifestation of borderline personality
symptoms. However, it was expected that the measurement of trait impulsivity would
remain fairly constant over time. Future prospective research among psychiatric inpatients
is needed to identify factors than can be predictive of future suicide attempts. It is also
important for future studies to collect information about specific pharmaceutical drugs or
“pills” used to attempt suicide among psychiatric patients, as this may aid in developing pre-
vention strategies.
It is also important to note that the clinical characteristics and risk factors in this study
apply to psychiatric patients who attempted suicide. This may not be generalizable to psychiat-
ric patients at risk for completed suicide, as differences in risk factors between those who
attempt and complete suicide have been reported in the literature [42]. Studies comparing risk
factors for completed suicide, specifically within psychiatric patient populations, are warranted
and can be used in conjunction with suicide attempt risk factors to develop an overall frame-
work for suicide prevention among patients with psychiatric disorders [43].
Conclusions
Findings from this study indicate that those who attempt suicide may not differ significantly
from psychiatric inpatients on sociodemographic factors such as unemployment, single mari-
tal status and living alone. Impulsivity and borderline personality symptoms were found to be
risk factors for suicide in this sample of psychiatric patients. A descriptive assessment of sui-
cide attempts indicated that limiting access to methods of suicide and educating social sup-
porters of hospitalized psychiatric patients regarding behavioural trends may be effective
suicide prevention strategies. Future qualitative research can identify themes associated with
thought processes leading to suicide attempts and feelings related to recovery among high-risk
psychiatric patients. Additionally, cohort studies among patients with serious psychiatric dis-
orders (i.e. psychiatric inpatients) are needed to establish the temporal association between
personality variables and attempted suicide and to further identify unique risk factors for sui-
cide attempts in psychiatric populations.
Supporting information
S1 File. Dataset.
(XLSX)
Acknowledgments
This work was supported by the Brain and Behavior Research Foundation Young Investigator
Grant (# 19058). The Brain and Behavior Research Foundation has no role in the design of the
study or publication of the results.
Author Contributions
Conceptualization: Meha Bhatt, Lehana Thabane, Zainab Samaan.
Data curation: Meha Bhatt, Rebecca B. Eisen, Wala El-Sheikh, Jane DeJesus, Sumathy Rangar-
ajan, Pam Mackie, Shofiqul Islam, Mahshid Dehghan.
Formal analysis: Meha Bhatt, Stefan Perera.
Funding acquisition: Lehana Thabane, Zainab Samaan.
Investigation: Rebecca B. Eisen, Sharon Yeung, Wala El-Sheikh, Jane DeJesus, Sumathy Ran-
garajan, Heather Sholer, Elizabeth Iordan, Pam Mackie, Shofiqul Islam, Mahshid Dehghan.
Methodology: Meha Bhatt, Stefan Perera, Laura Zielinski, Lehana Thabane, Zainab Samaan.
Supervision: Zainab Samaan.
Validation: Sharon Yeung.
Visualization: Meha Bhatt, Laura Zielinski.
Writing – original draft: Meha Bhatt.
Writing – review & editing: Meha Bhatt, Stefan Perera, Laura Zielinski, Rebecca B. Eisen, Sha-
ron Yeung, Wala El-Sheikh, Jane DeJesus, Sumathy Rangarajan, Heather Sholer, Elizabeth
Iordan, Pam Mackie, Shofiqul Islam, Mahshid Dehghan, Lehana Thabane, Zainab Samaan.
References
1. Sher L. Preventing suicide. Qjm. 2004; 97(10):677–80. https://doi.org/10.1093/qjmed/hch106 PMID:
15367739
2. Sher L. Teaching medical professionals about suicide prevention: what’s missing? Qjm. 2011; 104
(11):1005–8. https://doi.org/10.1093/qjmed/hcr125 PMID: 21824917
3. World Health Organization. Preventing suicide: A global imperative: World Health Organization. 2014.
http://www.who.int/mental_health/suicide-prevention/world_report_2014/en/
4. Mann JJ. Neurobiology of suicidal behaviour. Nat Rev Neurosci. 2003; 4(10):819–28. https://doi.org/10.
1038/nrn1220 PMID: 14523381
5. Centers for Disease Control and Prevention. Definitions: Self-directed violence. Injury prevention and
control: Division of violence. 2015. http://www.cdc.gov/violenceprevention/suicide/definitions.html.
6. Crosby A, Gfroerer J, Han B, Ortega L, Parks SE. Suicidal Thoughts and Behaviors Among Adults
Aged18 Years—United States, 2008–2009: US Department of Health and Human Services, Centers
for Disease Control and Prevention; 2011.
7. Goldman-Mellor SJ, Caspi A, Harrington H, Hogan S, Nada-Raja S, Poulton R, et al. Suicide attempt in
young people: a signal for long-term health care and social needs. JAMA psychiatry. 2014; 71(2):119–
27. https://doi.org/10.1001/jamapsychiatry.2013.2803 PMID: 24306041
8. Nock MK, Borges G, Bromet EJ, Alonso J, Angermeyer M, Beautrais A, et al. Cross-national prevalence
and risk factors for suicidal ideation, plans and attempts. The British Journal of Psychiatry. 2008; 192
(2):98–105. https://doi.org/10.1192/bjp.bp.107.040113 PMID: 18245022
32. Grossman DC, Mueller BA, Riedy C, Dowd MD, Villaveces A, Prodzinski J, et al. Gun storage practices
and risk of youth suicide and unintentional firearm injuries. Jama. 2005; 293(6):707–14. https://doi.org/
10.1001/jama.293.6.707 PMID: 15701912
33. Shah S, Hoffman RE, Wake L, Marine WM. Adolescent suicide and household access to firearms in
Colorado: results of a case-control study. Journal of Adolescent Health. 2000; 26(3):157–63. PMID:
10706163
34. Howlett A, Cheung AH. Substances used in completed suicide by overdose in Toronto: an observational
study of coroner’s data. Canadian journal of psychiatry. 2012; 57(3):184. https://doi.org/10.1177/
070674371205700308 PMID: 22398005
35. Leenaars AA, De Wilde EJ, Wenckstern S, Kral M. Suicide notes of adolescents: A life-span compari-
son. Canadian Journal of Behavioural Science/Revue canadienne des sciences du comportement.
2001; 33(1):47.
36. Garvey MJ, Spoden F. Suicide attempts in antisocial personality disorder. Comprehensive Psychiatry.
1980; 21(2):146–9. PMID: 7379506
37. Brodsky BS, Malone KM, Ellis SP, Dulit RA, Mann JJ. Characteristics of borderline personality disorder
associated with suicidal behavior. American Journal of Psychiatry. 1997.
38. Yen S, Shea T, Pagano M, Sanislow CA, Grilo CM, McGlashan TH, et al. " Axis I and Axis II disorders
as predictors of prospective suicide attempts: Findings from the Collaborative Longitudinal Personality
Disorders Study": Correction to Yen et al (2003). 2004.
39. Corruble E, Damy C, Guelfi J. Impulsivity: a relevant dimension in depression regarding suicide
attempts? Journal of affective disorders. 1999; 53(3):211–5. PMID: 10404706
40. Swann AC, Dougherty DM, Pazzaglia PJ, Pham M, Steinberg JL, Moeller FG. Increased impulsivity
associated with severity of suicide attempt history in patients with bipolar disorder. American Journal of
Psychiatry. 2014.
41. Baca–Garcia E, Diaz–Sastre C, Resa EG, Blasco H, Conesa DB, Oquendo MA, et al. Suicide attempts
and impulsivity. European archives of psychiatry and clinical neuroscience. 2005; 255(2):152–6. https://
doi.org/10.1007/s00406-004-0549-3 PMID: 15549343
42. Dorpat TL, Ripley HS. The relationship between attempted suicide and committed suicide. Comprehen-
sive Psychiatry 1967, 8(2):74–79. PMID: 6040467
43. Mościcki EK. Epidemiology of completed and attempted suicide: toward a framework for prevention.
Clinical Neuroscience Research 2001, 1(5):310–323.